Gene expression profiling in lung fibroblasts reveals new players in alveolarization

Olivier Boucherat, Marie-Laure Franco-Montoya, Christelle Thibault, Roberto Incitti, Bernadette Chailley-Heu, Christophe Delacourt, Jacques R. Bourbon


Little is known about the molecular basis of lung alveolarization. We used a microarray profiling strategy to identify novel genes that may regulate the secondary septation process. Rat lung fibroblasts were extemporaneously isolated on postnatal days 2, 7, and 21, i.e., before, during, and after septation, respectively. Total RNA was extracted, and cRNAs were hybridized to Affymetrix rat genome 230 2.0 microarrays. Expression levels of a selection of genes were confirmed by real-time PCR. In addition to genes already known to be upregulated during alveolarization including drebrin, midkine, Fgfr3, and Fgfr4, the study allowed us to identify two remarkable groups of genes with opposite profiles, i.e., gathering genes either transiently up- or downregulated on day 7. The former group includes the transcription factors retinoic acid receptor (RXR)-γ and homeobox (Hox) a2, a4, and a5 and genes involved in Wnt signaling (Wnt5a, Fzd1, and Ndp); the latter group includes the extracellular matrix components Comp and Opn and the signal molecule Slfn4. Profiling in whole lung from fetal life to adulthood confirmed that changes were specific for alveolarization. Two treatments that arrest septation, hyperoxia and dexamethasone, inhibited the expression of genes that are upregulated during alveolarization and conversely enhanced that of genes weakly expressed during alveolarization and upregulated thereafter. The possible roles of these genes in secondary septation are discussed. Gene expression profiling analysis on freshly isolated cells represents a powerful approach to provide new information about differential regulation of genes during alveolarization and pathways potentially involved in the pathogenesis of bronchopulmonary dysplasia.

  • bronchopulmonary dysplasia
  • hyperoxia
  • Hox
  • Wnt
  • Opn

new bronchopulmonary dysplasia (BPD) is a chronic lung disease of preterm infants, most often consecutive to respiratory distress. BPD is featured by disruption of the normal process of alveolarization (or alveologenesis), leading to enlarged and simplified air spaces (26). Although BPD represents a major health problem, the mechanisms that inhibit distal lung growth are poorly understood (6). Establishing a new basis for safe and effective intervention to prevent or reverse arrested alveolarization represents a pressing need. This can be achieved only by a better understanding of the mechanisms that regulate alveolarization.

Compared with earlier periods of lung development, particularly branching morphogenesis, the secondary septation process through which alveolar sacs are subdivided into definitive alveoli has been the object of a smaller number of studies. Myofibroblasts express smooth muscle actin (α-SMA) and synthesize elastin that forms deposits in the thickness of the walls of terminal air sacs. At the level of elastic fibers, crests surge that progressively elongate to give rise to secondary septa (10). This greatly increases gas exchange surface area and allows the animal to adapt successfully to extrauterine life. This process takes place postnatally for the major part in humans and is totally postnatal in rodents. To date, only a few key molecules involved in the process have been identified, including extracellular matrix (ECM) components, matrix-remodeling proteases, growth factors, and receptors (2, 5, 30, 42, 43, 47, 60). Four gene expression profiling studies have been performed recently to identify new factors involved in lung development (4, 12, 14, 35). Although these studies identified a useful list of differentially regulated genes, only two studies (12, 14) specifically addressed the alveolar stage of development. Moreover, they dealt with limitations imposed by the whole organ approach, i.e., restriction in the extent of gene expression changes, alterations induced by changes in the proportion of various cell populations within the lung, and impossibility of associating changes with the responsible cell type(s).

Because the formation of alveoli occurs in a strictly defined period, we hypothesized that genes that regulate alveolar development must be differentially expressed during the period of active alveolar septation compared with the preceding and following periods. In the rat, the formation of secondary alveolar septa takes place between the 4th and 14th postnatal days. Taking into account the above-emphasized pivotal role of interstitial cells, we searched for genes differentially expressed in rat lung fibroblasts isolated on postnatal days 2, 7, and 21. This allowed us to identify numerous genes that undergo significant changes in expression during alveolarization.

Particular attention was paid to genes either upregulated between postnatal days 2 and 7 and downregulated between days 7 and 21 or, conversely, downregulated between days 2 and 7 and upregulated between days 7 and 21. Moreover, we investigated the expression of some of these genes in two classical models of arrested alveolar development. These models, namely the chronic exposure of neonatal rats to ≥95% oxygen or to the synthetic glucocorticoid dexamethasone (Dex), are widely used to explore the mechanisms of BPD (9, 40). The data presented here extend our knowledge about normal alveolar formation, and consequently they should help in understanding the molecular impact of pathological disturbances on the developing lung.


Animals and Lung Tissue

Dated pregnant Sprague-Dawley rats were purchased from Charles River (Saint Germain sur l'Arbresle, France). The day of mating was designated day 0 of gestation. Term is 22 days. Lung tissues were collected between fetal day 18 and postnatal day 21. Lung tissues from adult rats (8 wk of age) were also collected. Fetuses were retrieved by cesarean section from pentobarbital-anesthetized pregnant females. Lungs from fetuses and pups were either immediately frozen in liquid nitrogen and then kept at −80°C until RNA extraction or used for cell isolation. Procedures involving animals were in accordance with the rules of the Guide for Care and Use of Laboratory Animals. The authors have been granted a license from the French Ministry of Agriculture to conduct the animal research described here.

Lung Fibroblast Isolation

Rat lung fibroblasts were extemporaneously isolated on postnatal days 2 (saccular stage), 7 (alveolar stage, progressing secondary septation), and 21 (terminated secondary septation) as described previously (11). In brief, cells were dispersed by trypsin and collagenase and then plated in plastic petri dishes in MEM-10% FBS and allowed to adhere for 45 min. Nonadherent cells were removed, and the adherent fibroblasts were rinsed and immediately scraped and pelleted. This method has been reported to result in fibroblast preparations with >95% purity (8, 11). Freshly isolated cells were stored at −80°C until RNA extraction for gene expression studies, without any culture step.

Generation of cRNA “Target” and Chip Hybridization

Total RNA was extracted from isolated fibroblasts by the guanidinium isothiocyanate method (TRIzol reagent, Invitrogen, Cergy-Pontoise, France), followed by purification using RNeasy columns (Qiagen, Courtaboeuf, France). Three biological replicates from three different litters were prepared for each time point, i.e., a set of fibroblasts was isolated from pups from each of the three litters at each time point. Integrity and purity of RNA were checked by spectrophotometry and capillary electrophoresis with the Bioanalyser 2100 and RNA 6000 LabChip kit from Agilent Technologies (Palo Alto, CA). cDNAs were synthesized with the Superscript Choice system (Invitrogen). Biotin-labeled cRNAs were then synthesized with the Affymetrix IVT labeling kit (Affymetrix, Santa Clara, CA). After purification, 10 μg of fragmented cRNA was hybridized to the Affymetrix GeneChip Rat Genome 230 2.0 Array (31,042 probe sets including >28,000 rat genes), and the chips were automatically washed and stained with streptavidin-phycoerythrin and a fluidics station. Finally, arrays were scanned at 570 nm with a resolution of 1.56 μm/pixel with the Gene Chip Scanner 3000 from Affymetrix. Expression values were generated with Microarray Suite v5.0 (MAS5, Affymetrix). Each sample and hybridization experiment underwent a quality control evaluation, including the percentage of probe sets reliably detecting between 50% and 60% Present call and a 3′-5′ ratio of GAPDH gene <3 (Table 1).

View this table:
Table 1.

Evaluation of quality filter criteria in microarray chips

Microarray Data Analysis

Gene expression values.

This calculation was made under the R statistical environment (20) with the “affy” R package of Bioconductor (22). Raw CEL data were corrected for background with the background correction of the robust multichip averaging (RMA) algorithm (23), normalized with the qspline method (61), PM-corrected by using only perfect match, and summarized by the Li-Wong model (27). Raw CEL data were corrected for background by the rma2 implementation of the RMA background correction method (20), normalized with the qspline method (61), using only perfect match. Resulting data were summarized by the Li-Wong model (27).

Replicate reproducibility.

To assess reproducibility, we computed the correlation between the expression values of the three biological replicates of each time point. All values were above 0.97, which indicates good reproducibility.

Gene selection.

Stringent filtering criteria were used to identify genes whose expression level was significantly changed between two stages. We performed a Kruskal-Wallis analysis on the three time points and a comparative analysis using MAS5 for the pairwise comparisons day 7 vs. day 2 and day 21 vs. day 7, which are those of interest in the present paper. We selected genes that met all of the following requirements: 1) Kruskal-Wallis P value <0.05 for the three replicates of each time point, 2) mean of MAS5 fold change over the three replicates either ≥2 or ≤0.5 for both comparisons, and 3) MAS5 change P value <0.01 for pairwise comparisons.

For any of the above steps, except for the comparative analysis, we used the R affy package of Bioconductor (22). To enable the visualization and illustration of our analyses, we used TIGR MultiExperiment Viewer (MeV 2.2) software (52). Genes with significant fold change ratio were exported to MeV 2.2 for hierarchical clustering, using centered correlation and average linkage. Online Mendelian Inheritance in Man (OMIM; and UniProt ( were used to provide information on gene functions. The primary data are available at the Gene Expression Omnibus Database (, GEO accession no. GSE7491). OMIM and the Gene Expression Omnibus Database are provided in the public domain by the National Center for Biotechnology Information (Bethesda, MD).

In Vivo Treatments of Rat Pups

Elevated O2 exposure.

Rat pups and their dams were placed in Plexiglas exposure chambers (Charles River) and run in parallel with either >95% or 21% (room air) inspired O2 fraction (FiO2) from day 0 to day 6. O2 concentrations were monitored regularly. Because adult rats have limited resistance to high O2, the dams were exchanged daily between O2-exposed and room air-exposed litters. Pups from two different litters were mixed so that there were littermates in both conditions. Chambers were opened for 20 min every day to switch dams between air and O2 environments and to clean cages. After the 6-day 95% O2 exposure period, some of the pups were allowed to recover in room air for 4 days.

Dexamethasone treatment.

Two different litters were subdivided into three groups that received an intraperitoneal injection of either 0.1 or 0.5 μg·g−1·day−1 water-soluble Dex (Sigma, L'Isle d'Abeau, France) or vehicle alone (saline, control group). Six animals were used for each experimental condition. Pups were killed on the day that followed the last injection of Dex or saline.

Collection of lung samples.

On day 6 or 10, rat pups were killed by an intraperitoneal overdose of pentobarbital sodium (70 mg/kg, Ceva, Libourne, France) and were exsanguinated by aortic transection. Lungs were either fixed at 20 cmH2O constant pressure for morphological analysis or immediately dropped in liquid nitrogen and kept frozen at −80°C until further RNA or protein extraction.

Real-Time Quantitative Polymerase Chain Reaction

RNAs from each extraction sample were reverse-transcribed into cDNAs, using 2 μg of total RNA, Superscript II reverse transcriptase, and random hexamer primers (Invitrogen) according to the supplier's protocol. Real-time PCR was performed on an ABI Prism 7000 device (Applied Biosystems, Courtaboeuf, France) with the following protocol: initial denaturation (10 min at 95°C) and then a two-step amplification program (15 s at 95°C followed by 1 min at 60°C) repeated 40 times. Melt curve analysis was used to check that a single specific amplified product was generated. Reaction mixtures consisted of 25 ng of cDNA, SYBR Green 2× PCR Master Mix (Applied Biosystems), and forward and reverse primers (Table 2) in a reaction volume of 25 μl. Primers were designed with Primer Express software (Applied Biosystems). Real-time quantification was monitored by measuring the increase in fluorescence caused by the binding of SYBR Green dye to double-stranded DNA at the end of each amplification cycle. Relative expression was determined by using the ΔΔCt (threshold cycle) method of normalized samples (ΔCt) in relation to the expression of a calibrator sample, according to the manufacturer's protocol. Each PCR run included a no-template control and a sample without reverse transcriptase. All measurements were performed in triplicate.

View this table:
Table 2.

Primer sequences for qPCR

Immunoblot Analysis

Rat lung tissue was homogenized in RIPA buffer containing protease inhibitors (Roche Diagnostics, Mannheim, Germany). Protein content was assayed by Bradford assay. Eighty micrograms of total proteins were electrophoresed on 12% SDS-polyacrylamide gels and then transferred onto polyvinylidene fluoride membrane (Millipore, Saint-Quentin en Yvelines, France). To document equivalent protein loading, membranes were stained with Ponceau S dye (Sigma) and photographed before antibody incubations. After blocking with 5% nonfat dry milk in Tris-buffered saline containing 0.05% Tween 20 (TTBS) at room temperature for 2 h, membranes were exposed overnight at 4°C to one of the following antibodies: mouse anti-osteopontin (Santa Cruz Biotechnology, Santa Cruz, CA), goat anti-schlafen4 (Santa Cruz Biotechnology), goat anti-norrin (R&D Systems), or rabbit anti-RXRγ (AbCys, Paris, France), all diluted in 2% nonfat dry milk in TTBS. After five rinses in TTBS, the membranes were incubated for 1 h with the appropriate secondary IgG peroxidase-conjugated antibody. Membranes were then incubated for 1 min in chemiluminescent detection reagent (ECL, GE Healthcare Life Sciences, Velizy, France) before exposure to KODAK BioMax MS film for 2 min.


Isolated cells adherent to plastic were methanol fixed at −20°C and immunostained with goat anti-vimentin (Santa Cruz Biotechnology), mouse monoclonal anti-pan-cytokeratin (Sigma), goat anti-CD31 (Santa Cruz Biotechnology), or mouse monoclonal anti-CD68/ED1 (Abcam, Paris, France) antibodies. Secondary antibodies were donkey anti-goat Alexa Fluor 594 and donkey anti-mouse Alexa Fluor 488 (Invitrogen). Simultaneous nuclear labeling was obtained with Vectashield mounting medium with DAPI (Vector Laboratories, Burlingame, CA). Immunostaining of postnatal day 10 lung tissue was performed with 5-μm paraffin sections that were deparaffinized and rehydrated. For epitope retrieval, the sections were boiled in 10 mM sodium citrate for 20 min and cooled for 20 min. Endogenous peroxidase activity was quenched for 15 min in a bath of 3% H2O2 in H2O. All sections were incubated with 2.5% normal horse serum for 30 min and then with goat anti-norrin (R&D Systems), rabbit anti-frizzled1 (MBL International, Woburn, MA), goat anti-Wnt5a, goat anti-Hoxa5 (Santa Cruz Biotechnology), or rabbit anti-RXRγ (AbCys) antibodies overnight at 4°C in a humidified chamber. After PBS washing, the sections were incubated for 15 min with horse biotinylated secondary antibody. After an extensive wash with PBS, all the sections were exposed to a streptavidin-peroxidase preformed complex for 15 min. Horse serum, biotinylated secondary antibody, and streptavidin-peroxidase were from the Vectastain Universal Quick Kit (Vector Laboratories). Finally, sections were covered with diaminobenzidine tetrahydrochloride (Vector Laboratories) for 2 min, counterstained with methyl green, dehydrated, and observed with a light microscope. As negative controls, primary antibodies were omitted.

Statistical Analysis of Whole Lung Studies

Data are presented as means ± SE. Multiple-group comparisons were made by nonparametric Kruskal-Wallis analysis, and two-group comparisons were made by nonparametric Mann-Whitney U-test. P = 0.05 was considered as the limit of statistical significance.


Microarray Analysis Identifies Gene Expression Patterns Associated With Alveolarization

In an attempt to identify developmentally regulated genes whose expression changes are associated with lung alveolarization, microarray transcription profiles were obtained from lung fibroblasts isolated at postnatal days 2, 7, and 21. Fibroblast enrichment of cell preparations was checked by immunolabeling for vimentin (fibroblast marker), cytokeratins (epithelial markers), CD31 (endothelial marker), and CD68/ED1 (macrophage marker). As shown in Fig. 1A, almost all cells were vimentin positive, indicating their interstitial cell character. Only a few cells stained for cytokeratin (Fig. 1B). Rare cells displayed very faint CD31 labeling (not shown). No CD68/ED1 labeling was detected. The proportion of genes designated as “Present” in isolated fibroblasts by the Affymetrix software was 57.0 ± 1.27%, 57.7 ± 0.92%, and 53.9 ± 1.62% at postnatal days 2, 7, and 21, respectively, and did not fluctuate significantly (Table 1). Based on the hypothesis that genes that regulate alveolar development must be differentially expressed during the period of active septum formation compared with preceding and following periods, we focused our analysis on genes simultaneously increased between days 2 and 7 and decreased between days 7 and 21 (genes designated upregulated during septation, Fig. 2) or, conversely, simultaneously decreased between days 2 and 7 and increased between days 7 and 21 (genes designated downregulated during septation, Fig. 3).

Fig. 1.

Immunostaining of cells isolated on postnatal day 7. Nuclear labeling was obtained with DAPI (blue fluorescence). Almost all the cells isolated from enzymatically dispersed lung cells after 45-min adherence to plastic were vimentin positive (red fluorescence in A). Rare cells were cytokeratin positive (green fluorescence, arrow in B).

Fig. 2.

Clustergrams of genes upregulated in fibroblasts during alveolar septation. Genes selected on the basis of significant change by MAS (see criteria in materials and methods) were exported to the TIGR MeV array tool for hierarchical clustering using Manhattan correlation-based distance and average linkage. Each row represents 1 particular gene, and each column represents expression ratio of that gene between 2 samples from the same litter at different stages. Data are presented as groups of 3 comparisons: 7-day lung fibroblasts compared with 2-day lung fibroblasts (first 3 columns) and 21-day lung fibroblasts compared with 7-day lung fibroblasts (last three columns). Genes that were present at higher levels in the examined group are shown in progressively brighter shades of yellow, whereas genes that were expressed at lower levels are shown in progressively brighter shades of blue. Genes in black were not different between the 2 groups. Gene symbols, gi accession numbers, and GO molecular function are listed adjacent to each gene. Expressed sequence tags were not included in the list. ND, not determined.

Fig. 3.

Clustergrams of genes downregulated in fibroblasts during alveolar septation. Symbols same as in Fig. 2.

Validation of Microarray Data

To validate microarray data with independent methods, we selected 13 genes and analyzed their expression by real-time quantitative polymerase chain reaction (qPCR), using the same samples as in the microarray experiment. Among upregulated genes, we selected midkine (MK) because of its previously known lung expression pattern (38) and homeobox genes (Hox) a2, a4, and a5 as well as the retinoic acid (RA) X receptor-γ (RXRγ) for their important role in embryogenesis and their potential contribution to the transcriptional regulation of genes involved in secondary septation. Frizzled-1 (Fzd1), wingless-type MMTV integration site 5a (Wnt5a), and Norrie disease protein or norrin (Ndp) were also retained for validation because of their prominent roles in developmental events during embryogenesis (31). Among downregulated genes, we selected ECM components or cell matrix adhesion molecules, including tenascin-X (TnX), cartilage-oligomeric matrix protein (Comp), osteopontin (Opn), and osteoactivin (Gpnmb). Finally, a member of the schlafen (Slfn) protein family implicated in the regulation of cell growth, Slfn4 (7), was selected because its expression in the lung had not been reported previously. For each selected gene, the fold change determined by qPCR was similar to or higher than that determined by the microarray analysis (Table 3), indicating a high degree of concordance between results from both methods. This suggests a high degree of confidence in our overall microarray data.

View this table:
Table 3.

Comparison of fold change ratios obtained by arrays and qPCR

Expression Profile of Selected Genes From Fetal Life to Adulthood

To determine whether 1) changes in isolated fibroblasts were detectable also in whole lung homogenate and 2) observed changes were specific of alveolarization, we evaluated the mRNA level of selected genes in the developing rat lung from fetal day 18 (canalicular stage) to adulthood. Genes of the first group, i.e., those upregulated in fibroblasts, showed quite similar expression patterns although with variable amplitude (Fig. 4, A and B). The expression of all three Hox gene and RXRγ changed little until postnatal day 3 and then increased two- to fourfold on day 5, remained elevated during alveolar septation, and returned to prenatal levels on day 20 to further decrease to extremely low levels in adulthood (Fig. 4A). The expression profiles of Wnt5a, Fzd1, and Ndp were similar to those of transcription factors with sustained elevated expression between days 5 and 14 of about 2.5-, 3.5-, and 6.5-fold the day 3 level, respectively, and marked downregulation on day 20 (Fig. 4B). In the second group, i.e., genes downregulated in fibroblasts (Fig. 4C), Slfn4 and Gpnmb expression peaked on fetal day 21 and postnatal day 1, respectively (10- to 12-fold increase compared with fetal day 18), decreased 2- to 3-fold during alveolar septation, and then increased again 3- and 1.5-fold, respectively, after its completion; the level of Sfln4 then stayed elevated, whereas the level of Gpnmb dropped in the adult lung (Fig. 4C). TnX expression stayed unchanged from fetal day 18 to postnatal day 3, increased slightly on day 5 and then sharply on days 14–20, when it reached about 12 times the fetal and postnatal level, and returned to a very low level in the adult lung. These three genes therefore share the feature of peaking during the second phase of alveolarization that follows the completion of septation.

Fig. 4.

Quantitative developmental changes of mRNA expression level in whole lung tissue of a selection of genes shown to be up- or downregulated during septation in the microarray study. Expression was quantified by real-time PCR from fetal life (canalicular stage of development) to adulthood in 3 individual lung samples per stage. Relative expression was referred to fetal day 18 (f18) level, arbitrarily set at 100. A: transcription factor genes found to be upregulated in fibroblasts during alveolar septation by microarray analysis. B: genes of signal molecules and receptor found to be upregulated in fibroblasts during alveolar septation by microarray analysis. C: genes found to be downregulated in fibroblasts during alveolar septation by microarray analysis. The study indicates that major changes were specifically associated with the period of alveolarization. Values are means ± SE; absence of an error bar indicates that the corresponding SE was too small to be represented at this scale. *P < 0.05 compared with the preceding stage (nonparametric comparison by Mann-Whitney test).

Localization of Upregulated Gene Expression in Postnatal Rat Lung

Immunostaining for WNT5A, NDP, FZD1, HOXa5, and RXRγ proteins was performed on sections from day 10 rat lungs. For each gene, immunostaining was detected in the thickness of septa, including at the tip of growing secondary septa (Fig. 5). In addition, a strong immunostaining for NDP and RXRγ was also observed in bronchial epithelial cells (Fig. 5, C and I).

Fig. 5.

Immunohistochemical localization of WNT5A (A, B), NDP (C, D), FZD1 (E, F), HOXA5 (G, H), and RXRγ (I, J) on formaldehyde-fixed sections of normal rat lungs aged 10 days. B, D, F, H, and J are enlargements of the dashed boxes in A, C, E, G, and I, respectively. Staining was found in both primary and secondary septa, including growing tips (arrows). A strong immunostaining for NDP and RXRγ was also observed in bronchial epithelial cells (arrowheads in C and I, respectively). Insets in solid-line boxes represent negative controls. Bar = 50 μm in A, C, E, G, and I and 20 μm in B, D, F, H, and J.

Effects of Dex on Selected Gene Expression in Postnatal Rat Lung

Consistent with previous studies (40, 51), Dex (0.1 μg·g−1·day−1) given from postnatal days 1 to 5 induced marked impairment of alveolar septation on postnatal day 6, with enlarged distal air spaces (Fig. 6B) compared with the normal alveolar structure of control littermates (Fig. 6A). As regards upregulated genes during septation, Hoxa5, Wnt5a, and Ndp mRNA levels showed a 40–50% decrease with this dosage, whereas no change was observed for Hoxa2 and a4 and Fzd1 (Fig. 6C). When Dex dosage was increased to 0.5 μg·g−1·day−1, the expression of all selected genes except RXRγ was decreased 40–80%. As regards downregulated genes, 0.1 and 0.5 μg·g−1·day−1 Dex caused a similar twofold increase in Gpnmb and Comp mRNA abundance compared with controls (Fig. 6D). The effect was stronger for Slfn4 and Opn mRNA abundance, with 4- and 11-fold increases, respectively. No change was observed for TnX expression level at any Dex dosage.

Fig. 6.

Effects of 0.1 or 0.5 μg·g−1·day−1 dexamethasone (Dex) on selected gene expression in postnatal rat lung. A and B: hematoxylin and eosin (H&E)-stained sections of lungs from 6-day-old rats treated daily with saline (A) or 0.1 μg/g Dex (B) since the day of birth. C and D: real-time PCR analysis of selected genes upregulated (C) and downregulated (D) during alveolarization, performed on whole lung of 6-day-old control and Dex-treated rats. Dex reduced the expression of upregulated genes and enhanced that of downregulated genes. Values are means ± SE of 6 individual samples per group; nonparametric Kruskal-Wallis multiple-group comparison and 2-group comparisons by Mann-Whitney U-test. *P < 0.05, #P < 0.01 for Dex vs. saline.

Effects of 95% O2 and Recovery in Air on Selected Gene Expression in Postnatal Rat Lung

Consistent with previous observations (9), newborn rats exposed to hyperoxia for 6 days displayed large, simplified air spaces (Fig. 7B) compared with the alveolar structure of littermate controls exposed to air (Fig. 7A). When O2-exposed rats were subsequently allowed to recover in room air for 4 days, normal lung structure caught up (Fig. 7D) compared with 10-day controls in air (Fig. 7C). The expression level of all upregulated genes (Fig. 7E) was decreased ∼50% under O2 compared with controls. Restoration to levels no longer significantly different from those in controls occurred for Hox genes, RXRγ, and Wnt5a after return in air for 4 days. By contrast, Fzd1 and Ndp mRNAs remained a significantly lower than normal level, although they increased significantly compared with day 6 under hyperoxia, indicating partial recovery. Similar to the Dex model, downregulated genes (Fig. 7F) displayed an opposite pattern. Comp and TnX mRNA levels were increased 2-fold by hyperoxia, and Gpnmb, Slfn4, and Opn mRNA levels were increased 5-, 6-, and 32-fold, respectively. After the 4-day recovery in air, mRNA levels were similar to those in air-exposed controls for all genes.

Fig. 7.

Effects of 95% O2 and recovery in air on selected gene expression in postnatal rat lung. A–D: H&E-stained sections of lungs from 6-day-old rats exposed to air (A) or to 95% O2 (B) and from 10-day-old rats exposed to air (C) or to 95% O2 (D) for 6 days and allowed to recover in room air for 4 days. E and F: real-time PCR analysis of selected genes upregulated (E) and downregulated (F) during alveolarization performed on whole lung of rat pups. Hyperoxia reduced the expression of upregulated genes and enhanced that of downregulated genes. Return to air for 4 days allowed partial or complete recovery to occur. G: effects of 95% O2 on RXRγ, NDP, OPN, and SLFN4 protein expression in postnatal rat lung. Changes in protein content consistent with changes in corresponding transcript were observed. Values are means ± SE of 5 individual samples per group; nonparametric Kruskal-Wallis multiple-group comparison and 3-group comparisons by Mann-Whitney U-test. *P < 0.05, #P < 0.01 for day 6 O2 or day 10 air vs. day 6 air; §P < 0.05 for day 6 O2 + day 4 air vs. day 10 air.

To determine whether changes in mRNA level were reflected in lung protein contents, we performed Western blots for RXRγ, NDP, OPN, and SLFN4, which displayed the largest variations of expression. As shown in Fig. 7G, protein changes were concordant with RNA changes, since RXRγ and NDP amounts were decreased in O2 and, reciprocally, OPN and SFLN4 amounts were increased.


Rationale, Limitations, and Validity of the Approach

Understanding the molecular mechanisms of alveolarization is important for the prevention and care of alveolar growth arrest in BPD as well as for maintenance and regeneration of the alveolar structure in emphysema. In the rat, alveolar formation is an entirely postnatal event that initiates on the fourth postnatal day and is completed by 3 wk (10). Alveolar number increases linearly over this period (50), but alveolar density reaches maximal value on day 8 and does not increase further thereafter (44). This is consistent with the concept of a two-phase model of alveolarization consisting of saccular subdivision by secondary crests until day 8 and acinar extension with alveoli added distally in a second step (39). Consequently, the choice of postnatal days 2, 7, and 21 for microarray analysis should allow us to discriminate between genes involved in secondary septation and those that are involved later. This is of particular interest considering that impaired secondary septation is the major feature of BPD.

Formation of secondary septa is dependent on harmonious interactions between several cell types. Because fibroblasts are key players in alveolar septation, we searched for genes differentially expressed in fibroblasts that were isolated at the three stages listed above. This, along with investigating the effects of treatments that impair septation, allowed us to identify genes the expression of which appears to have to be either enhanced or repressed for proper course of the process. Considering the cell complexity of the lung, cell type-specific gene profiling studies present several advantages for interpreting the results, including removing a lot of noise from the system. Although our approach did not allow detection of expression changes in epithelial and endothelial cells, which is clearly a limitation of the present study, this approach facilitated the interpretation of data. Importantly, a subsequent study in whole lung confirmed the significance of changes evidenced in fibroblasts. Moreover, our microarray data pointed out stage-associated changes in expression of some genes previously recognized to be involved in alveolarization. These include, for instance, midkine and drebrin. midkine expression was previously shown to be transiently increased in normal lungs between postnatal days 2 and 7, diminished by hyperoxia or glucocorticoids, and upregulated by RA (24, 41). Drebrin, an actin-binding protein, has been detected in cell processes of myofibroblasts during the maturation of alveolar septa (63). The developmental expression patterns previously reported in whole lung for these genes are very similar to those obtained here from isolated fibroblasts. Our microarray analysis also identified genes the targeted inactivation of which altered secondary septation, including Hoxa5, Fgfr3, and Fgfr4 (34, 49, 60). Together, these findings indicate that the transcriptome status of fibroblasts was unlikely to have been altered extensively by cell isolation, which validates the approach.

Nevertheless, the pulmonary fibroblast population is heterogeneous. Fibroblasts are present in vascular and airway adventitia, and in the alveolar area two subclasses known as myofibroblasts and lipofibroblasts (or lipid-laden fibroblasts) are present (56). Consequently, whole lung digest followed by adherence-based cell selection results in a somewhat heterogeneous population of fibroblastic cells. We therefore cannot exclude the possibility that a subset of regulated genes during alveolarization evidenced in the present study may finally prove not to be expressed in alveolar myofibroblasts.

Genes Upregulated During Septation

This study identified numerous genes that were upregulated in lung fibroblasts during secondary septation whose involvement in the process had not been suspected previously. The expression study in whole lung of a selection of these genes indicated that they returned to low levels after completion of the septation process, i.e., between postnatal days 14 and 20. Their protein products were localized in the thickness of growing septa. Together, these findings strongly argue in favor of their involvement in the alveolarization process. Interestingly, we found that in addition to Hoxa5 the mRNAs from two other Hox genes, Hoxa2 and Hoxa4, were increased during alveolarization and deeply diminished in animal models of arrested septation. Hox genes encode homeodomain transcription factors that are important for specifying tissue and cell type identities along the anterior-posterior axis of the embryo (33). They also have a variety of functions in the development of a number of organs. Hoxa5 inactivation leads to severe respiratory tract defects (3) and impairs alveolarization because of mispositioning of alveolar myofibroblasts (34). By contrast, previous investigations on both other Hox genes had not revealed involvement in lung alveolar development. Mice homozygous for a targeted mutation of the Hoxa2 gene are born with a bilateral cleft of the secondary palate associated with multiple head and cranial anomalies, and they die within 24 h of birth (17). Consequently, evaluation of the consequences of this deletion for secondary septation of the lung is not possible. With regard to Hoxa4, its expression was found in structures partially derived from the lateral plate or para-axial mesoderm, including the fetal lung, intestine, and kidney (16). Lack of lung phenotype during the perinatal period in Hoxa4-null mice may be due to functional redundancy among Hox genes. It should be emphasized that although the latter have been implicated in the regulation of a variety of pathways, few target genes have been demonstrated to be placed under their direct regulatory control (53). Hence, to fully evaluate the role of Hox proteins in alveolar development, future research requires identification of their target genes in addition to determination of their respective specific functions.

Among the transcription factors whose expression is increased during alveolarization, we also found a member of the RA pathway, namely RXRγ. There are two classes of RA-activated transcription factors, the RARs and RXRs, which form homo- and heterodimers and bind both all-trans-RA and 9-cis-RA or only 9-cis-RA, respectively. Each class possesses three isoforms, α, β, and γ (25). Exogenous and endogenous retinoids have been reported to positively influence alveolarization (15, 37). Consistently, the crucial role of several RA receptors for alveolar development has been demonstrated. For instance, mice deleted for both RARγ and RXRα genes fail to form alveoli correctly, whereas mice deleted for the RARβ gene form too many alveoli (38, 42). The expression of several RARs has been reported to be enhanced in mouse lung when alveolar septation starts (18). Thus far, however, RXRγ had not been involved. Its developmental increase in fibroblasts coincidental with septation suggests an important function in these cells during the process, possibly in partnership with other receptors. The orphan receptor NR4A1 and the thyroid hormone receptor (THRα), which also display maximal gene expression on day 7 (Fig. 2) and have previously been reported to interact with RXRγ, are potential partners. Identification of the role and targets of RXRγ therefore appears as an important objective in the understanding of alveolarization mechanisms.

Three genes involved in Wnt signaling also displayed a marked increase during secondary septation, including Wnt5a, Ndp, and Fzd1. Wnt factors and NDP are diffusible mediators that bind to frizzled transmembrane receptors (62). After binding to frizzled receptors, intracellular Wnt signaling occurs through canonical and noncanonical pathways to regulate a lot of processes during development (31). The importance of Wnt signaling during lung development has been demonstrated in several ways. A number of Wnt ligands and frizzled receptors are expressed dynamically during lung development (55). Moreover, lung deletion of β-catenin, an essential downstream effector of the canonical Wnt signaling, disturbs lung proximal/distal cell fate and leads to death from respiratory failure at birth (45). Few data are available specifically on Fzd1 and Wnt5a in lung development. Fzd1 was shown to be expressed in the developing lung mesenchyme, consistent with present findings, and in presumptive bronchial smooth muscle as well as in vascular smooth muscle tissues such as those of the aorta and large pulmonary vessels (58). It may mediate Wnt signaling originating from other lung cell type(s). In Wnt5a−/− mice, lung saccular development was delayed, with thickened septa and increased cell proliferation in both epithelium and mesenchyme (28). Wnt5a−/− neonates die shortly after birth, however, which prevents evaluation of the consequences for alveolarization. To date, no published data highlight the involvement of Ndp in lung development. Ndp-deficient mice, a Norrie disease animal model, develop blindness because of distinct failure in retinal angiogenesis and complete lack of the deep capillary layers of the retina. Taking into account on the one hand the deficit of capillary/alveolar apposition in Wnt5a−/− fetuses (28) and on the other hand that WNT5a and NDP enhance angiogenesis through frizzled-4 interaction (36, 62), the hypothesis that both mediators play a role in lung microvascular development appears likely. Microvascular development is known to be required for normal alveolarization and to be controlled by VEGF produced by epithelial cells (43). Moreover, expression of each of these genes was deeply depressed in animal models of impaired alveolarization (Figs. 6C and 7E). We hypothesize that WNT5a and NDP may mediate another angiogenic control mechanism through mesenchymal-endothelial cell interaction in the process of secondary septa formation.

Genes Downregulated During Septation and/or Enhanced After Its Completion

We found Slfn4, a negative regulator of cell proliferation (7), to be downregulated in the lung from days 1 through 8, i.e., a period characterized by extensive overall cell proliferation (19). This period was strikingly flanked by two expression peaks. Moreover, Slfn4 expression was increased in animal models of arrested alveolarization. This may account at least in part for the inhibitory effect of hyperoxia and Dex on lung cell proliferation (32, 59). Further work will be necessary to clarify Sfln4 function in vivo and its mechanism of action.

Several genes that share similar expression profile during alveolarization encode proteins that are integral parts of ECM. Evidence has accumulated over recent years for a crucial role of ECM components in alveolar development (6). However, none of those genes identified in the present study had previously been related to this process. Cartilage oligomeric matrix protein (Comp) is a member of the thrombospondin family also referred to as thrombospondin-5 (Tsp5). To date, there is limited evidence of Comp/Tsp5 expression outside the skeleton. Although several members of the thrombospondin family have the ability to inhibit angiogenesis (1), Comp does not share this feature in vitro (57). Its function during alveolarization can only be a matter of speculation. In addition to Comp, we found that Opn gene expression was diminished during the window of secondary septation and dramatically increased in BPD models. OPN is a secreted RDG-containing phosphoprotein with structural and functional characteristics of a matricellular protein that has been implicated in a number of physiological and pathological events, although this is often controversial. OPN exists both as an immobilized ECM molecule in mineralized tissues and as a cytokine-like soluble protein that mediates cellular functions involved in inflammation and ECM remodeling (13). During development, OPN is essential to mammary gland differentiation (46). Altered wound healing was observed in Opn-deficient mice (29), whereas recombinant OPN enhanced migration and adhesion of a murine fibroblast cell line (54). Furthermore, Opn-null mice developed more severe acute lung injury than wild-type mice, and their survival times were shorter than those of their matched wild-type controls (65). Finally, Opn is highly expressed in pulmonary fibrosis, a feature encountered to variable extent in BPD (48). These data suggest a protective role of OPN in vivo and physiological functions during matrix reorganization after injury. Further studies should aim at determining the significance of decreased expression before septation and the role of increased expression in arrested septation.

We also included in the genes of interest that of the ECM protein tenascin-X. Deficiency of the latter was described as the molecular basis of a recessive type of Ehlers-Danlos syndrome characterized by abnormalities in elastic fiber morphology (66). Consistently, a recent study suggested that TnX is important in derma for the stability and maintenance of established elastin fibers, rather than for the initial phase of elastogenesis (67). The striking upregulation of TnX gene expression at the end of alveolarization when elastin deposition is completed is therefore consistent with the assumption that TnX may play a similar role in alveolar septa.

Finally, the upregulation of these various genes between days 14 and 20 and their return to low expression levels in the adult lung are suggestive of an involvement in the last part of alveolarization, when septa get thinner and microvascular maturation takes place. Further studies should determine whether their changes and those of genes with similar expression profile are associated with these processes.

Gene Expression Balance/Imbalance and Effects on Septation

Strikingly, the expression of genes up- and downregulated during secondary septation was invariably decreased or enhanced, respectively, in models of arrested septation. We cannot rule out the possibility that these changes may be due at least in part to changes in the proportions of the various lung cell types induced by the treatments. Dex indeed reduces the volume of interstitial fibroblasts (40). However, whereas this may account for reduced expression of genes that are upregulated in fibroblasts during septation, it could hardly explain the considerable increase of those that are downregulated in these cells during the process. Changes induced by treatments are therefore more likely to be due to altered expression levels. Both glucocorticoid administration and exposure to hyperoxia similarly result in an arrest of alveolar development. Postnatal glucocorticoid treatment induces precocious microvascular lung maturation that is likely to prevent further septation (51) but indeed induces no inflammation, whereas inflammatory response contributes to the deleterious effect of high oxygen levels on the neonatal lung (64). Although Dex and hyperoxia are therefore likely to operate through quite different mechanisms, they presented very similar consequences for the expression of most of the selected genes. As a common consequence of both treatments, arrested alveolar septation strongly argues in favor of an important role of the selected genes in alveolar formation. Together with the characteristic developmental expression profiles of these different genes, this indicates that alveolar septation not only involves upregulation of specific genes but also necessitates downregulation of other sets of genes that are much more strongly expressed during periods that immediately precede and follow the septation period. Because corticosteroids are believed to terminate the process of secondary septation by inducing thinning of septa and fusion of capillary vessels (40, 51), Dex treatment is likely to induce precociously those events that occur later in normal development. This assumption is consistent with the inhibition of genes that are upregulated when alveolar septation initiates, and with enhanced expression of genes that are upregulated at the end of the process. Because of the antagonistic effects of glucocorticoids and retinoids (37), it is therefore tempting to speculate that, among other possible mechanisms, retinoids trigger the induction of upregulated genes and repress downregulated genes in a first step, whereas glucocorticoids repress the former and trigger the induction of the latter in a second step. Imbalance in these chronological regulations may precipitate the disorders that lead to impaired alveolarization in BPD. It is not possible from the present study to determine whether the same or different populations of fibroblasts are involved in either phase. Further investigations should address this question.

In conclusion, the present data identify novel genes that are likely to be involved in the regulation of secondary septation and suggest that alterations of their expression could account for some of the disorders that characterize BPD. Based on information obtained from this study, further investigations will aim at characterizing their functions. This might provide new therapeutic targets for the prevention and care of BPD.


This work was supported by a grant from the Réseau National des Génopoles (Project No. 212).

O. Boucherat was supported by a doctoral fellowship grant from the Fondation pour la Recherche Médicale.


  • Address for reprint requests and other correspondence: J. Bourbon, INSERM U841, Faculté de Médecine, 8 rue du Général Sarrail, 94000 Créteil, France (e-mail: jacques.bourbon{at}

    Article published online before print. See web site for date of publication (


View Abstract